AutoMaker-Org/automaker

Automaker is a desktop and web application that orchestrates AI agents to autonomously implement software features based on descriptions you provide on a Kanban board.

The tool addresses the friction of manual code writing by letting developers describe features at a high level while AI agents handle implementation details. You define tasks on a Kanban board, approve plans generated by the agents, and watch as they execute the work across your codebase. The system uses git worktree isolation to keep agent work separate, supports real-time streaming of agent activity, and coordinates multiple agents on complex tasks. The underlying engine is powered by Claude Agent SDK, though the tool supports other providers including Codex, Copilot, Cursor, Gemini, and OpenCode.

Automaker suits teams and individual developers who want to accelerate feature development by delegating implementation to AI agents rather than writing code manually. It works best for projects where you can clearly articulate requirements and accept that AI-generated code requires review before merging. The tool is particularly valuable if you already use AI-assisted development workflows and want a more structured, agent-orchestrated approach. The README emphasizes that Automaker itself was built using agentic coding techniques, demonstrating the methodology it enables.

The project shows active development with a monorepo structure spanning frontend, backend, and shared libraries. The codebase is written primarily in TypeScript and built with React, Vite, Electron, and Express, indicating a mature technical foundation. The team maintains comprehensive documentation covering architecture, deployment, testing, and environment configuration, and provides both interactive TUI and graphical interfaces for users. Development includes testing infrastructure, linting standards, and production build processes, suggesting sustained engineering discipline.